文献类型：会议论文

英文题名：Unexpected Phenomenon: LLMs' Spurious Associations in Information Extraction

作者：Zhang, Weiyan[1];Lu, Wanpeng[1];Wang, Jiacheng[1];Wang, Yating[1];Chen, Lihan[2];Jiang, Haiyun[3];Liu, Jingping[1];Ruan, Tong[1]

机构：[1]East China Univ Sci & Technol, Sch Informat Sci & Engn, Shanghai, Peoples R China;[2]Beijing Inst Control Engn, Beijing, Peoples R China;[3]Tencent AI Lab, Shenzhen, Peoples R China

会议论文集：62nd Annual Meeting of the Association-for-Computational-Linguistics (ACL) / Student Research Workshop (SRW)

会议日期：AUG 11-16, 2024

会议地点：Bangkok, THAILAND

语种：英文

摘要：Information extraction plays a critical role in natural language processing. When applying large language models (LLMs) to this domain, we discover an unexpected phenomenon: LLMs' spurious associations. In tasks such as relation extraction, LLMs can accurately identify entity pairs, even if the given relation (label) is semantically unrelated to the pre-defined original one. To find these labels, we design two strategies in this study, including forward label extension and backward label validation. We also leverage the extended labels to improve model performance. Our comprehensive experiments show that spurious associations occur consistently in both Chinese and English datasets across various LLM sizes. Moreover, the use of extended labels significantly enhances LLM performance in information extraction tasks. Remarkably, there is a performance increase of 9.55%, 11.42%, and 21.27% in F1 scores on the SciERC, ACE05, and DuEE datasets, respectively.